MULTICOMPONENT

DISTILLATION

CHE-402
Date: 25/09/2018
Lecture No. 4
Some Additional Terminology

 When dealing with multi-component systems, some new

terminology in addition to the terms used in binary distillation
has been introduced
 Key & Non-key components
 Splits – distributing and
non-distributing systems
 Fractional recoveries

2
Key Components

 The components that have their distillate and bottoms

composition specified are known as the key
components.

 The most volatile of the key components is termed the

light key (LK).

 The least volatile of the key components is termed the

heavy key (HK).

3
Non-Key Components
 All other components not specified in the distillate or botoms are
termed non-key components (NK’s).

 If a non-key component is more volatile than the light key, then it

is termed a light non-key (LNK).

 If a non-key component is less volatile than the heavy key, it is a

heavy non-key (HNK).

 If a non-key component is neither a heavy non-key nor a light non-

key, then it is an intermediate non-key (INK) or simply NK.

4
Non-Key Component Splits

 The split of the non–key components is generally

defined as to where the non–key components are
obtained with respect to the distillate or bottoms
stream.

 One can have two types of situations concerning the

split of the non–key components:
 Sharp split – Non-distribution of non-keys
 Split – Distribution of non-keys

5
Non-distribution of NK’s

 Non–distribution of non–keys means that essentially

all of the non–keys are obtained in either the
distillate stream or the bottoms stream. We obtain a
sharp split of the NK’s.

 Non–distribution of non–keys can be assumed when:

 All of the non-keys are either HNK’s or LNK’s
 The fractional recoveries of the LK in the distillate
and HK in the bottoms are relatively large.

6
Distribution of NK’s

 Distribution of non–keys means that the non-keys

are not sharply split between the distillate stream or
the bottoms stream. We obtain a split of the NK’s.

 Distribution of non–keys occurs when:

 Not all of the non-keys are either HNK’s or LNK’s –
we have NK’s.
 The fractional recoveries of the LK in the distillate
and HK in the bottoms are not relatively large.

7
How do we determine the keys and the non–
keys in MCD?

 The classification of components in MCD can be determined from their relative

volatilities.

 Relative volatility is defined as the ratio of the K values for two components,
which is trivial for a binary system.

 In order to use relative volatilities in MCD, we choose a reference component

and define all other component volatilities with respect to the reference
component.

 The relative volatility for the reference component, of course, will be 1.

 We can then define relative volatilities using equilibrium coefficient K values for
each component, e.g., from the DePriester charts for hydrocarbon systems.

 The choice of the reference component depends upon the problem, but in
general it will be the HK component since it is less volatile than the LK
component.

8
Key and Non-Key Example
 Consider a distillation column with the following feed
components:
 propane
 n–butane
 n–pentane
 n–hexane

 The recoveries for n–butane and n–pentane are specified for

the distillation.

 What are the key and non–key designations for this separation?

9
Key and Non-Key Example
 We have:
Volatilities
propane > n–butane > n–pentane > n–hexane

Since the recoveries of n–butane and n–pentane are specified…

Component Designation
Propane Light Non–Key
n–butane Light Key
n–pentane Heavy Key
n–hexane Heavy Non–Key
10
Key and Non-Key Example
 If the recoveries of n-butane and n-hexane are
specified:

Volatilities
propane > n-butane > n-pentane > n-hexane

Component Designation
Propane Light Non-Key
n-butane Light Key
n-pentane Non-Key
n-hexane Heavy Key

11
Key and Non-Key Example
 If only the recovery of n–butane is specified:
Volatilities
propane > n–butane > n–pentane > n–hexane

Component Designation
Propane Light Non–Key
n–butane Key
n–pentane Non–Key
n–hexane Non–Key

12
Missing Keys
 In typical MCD problems, LK and the HK recoveries are specified.

 The fractional recovery of the missing key needs to be determined, but it must
be estimated since not enough information is typically given in the problem to
determine it directly.

 One way to estimate the fractional recovery of the missing key component is to
do an external mass balance based upon a binary system comprised of the LK
and HK.

 This fractional recovery is then used in the MCD solution. A trail–and–error

solution may be required to determine the actual fractional recovery.

13
Fractional Recoveries
 A fractional recovery, fi , is the amount or flow rate of component i in the
distillate or bottoms stream with respect to the amount or flow rate of
component i in the feed stream:

Dx i dist Dx D, i
 fi dist  1   fi bot
 f i dist  
Fz F, i Fz F, i
Wx i bot Wx w, i
 f i bot    fi bot  1   fi dist
Fz F, i Fz F, i

 It is the simple relationships expressed by the right-hand-side equations

that make the use of fractional recoveries useful.

 These are also often specified simply as % recovery.

14
Approximate Shortcut Methods
for Multicomponent Distillation

 Fenske equation
 calculation of multicomponent separation at
total reflux

 Underwood equations
 calculate minimum reflux ratio

15
 Approximate Shortcut Methods
for Multicomponent Distillation

 Estimate number of equilibrium

stages
 using an empirical correlation
 relates actual number of stages to :
1. number of stages at total reflux
2. minimum reflux ratio
3. actual reflux ratio

 Estimate feed location with an

empirical correlation
16
Fenske Equation - Total
Reflux
 Assumes the stages are equilibrium stages

 x A   xA  
log     
 x B  dist  x B  bot  (Eq. 15.3)
N min =
log ( AB) avg

17
Fenske Equation - Total
Reflux

 Dx A   Wx A  
log     
 Dx B  dist  Wx B  bot 
N min =
log ( AB) avg

18
Fenske Equation - Total
Reflux


(Eq. 15.4)

19
Fenske Equation - Total
Reflux
 Feasible alternatives:

1. If assume all non-keys are non

distributing:

 DxLNK , dist = FzLNK and xLNK , bot = 0

 BxHNK , bot = FzHNK and xHNK , dist = 0

20
 Fenske Equation - Total
Reflux
 Using these equations, the LK and
HK mass balance equations, and

 (Dx
i
i , dist )=D

 (Wx
i
i , bot )=W

 Distillate and bottom compositions

can be found
21
Fenske Equation - Total
Reflux
 Find Nmin by applying the Fenske
equation to the key components

 Check the assumption of non-

distribution of the non-keys by
fractional recovery equation for
non-keys

22
Fenske Equation - Total
Reflux
 If original assumption is invalid
 calculated values for non-key
compositions can be used to calculate
LK and HK compositions in distillate
and bottoms

 then apply Fenske equation again

23
Fenske Equation - Total
Reflux
2. If non-keys do distribute:

 Reasonable 1st guess for distribution

is required
 Guess can be obtained by assuming the distribution
of non-keys is the same at total reflux as it is at
minimum reflux.
 Distribution at minimum reflux obtained from
Underwood equation.

24
 Minimum Reflux

 Minimum reflux corresponds to infinite number of plates where the

operating line intersects or touches (pinches) the equilibrium curve.
In binary systems only one pinch point occurs. Either at feed point
or due to non-ideal behavior of mixtures, it can be either in
rectifying section or in stripping section
Near the pinch point,
 composition changes little from stage to stage
 - passing streams are very close to equilibrium

25
 For Binary System

 Figure a : For ideal mixture distillation at minimum reflux, most of the stages are crowded
into a constant-composition zone that bridges the feed stage. In this zone, all vapor and
liquid streams have compositions essentially identical to those of the flashed feed. This
zone constitutes a single pinch point.
 Figure b: If non ideal phase conditions create a point of tangency between the equilibrium
curve and the operating line in the rectifying section, the pinch point occurs within the
rectifying section. Alternatively, the single pinch may occur in the stripping section.

26
 For Multicomponent System
Class 1 Class 2

 For Class 1 separations, all feed components distribute to both the distillate and bottoms
product and a single pinch point bridges the feed stage (Figure c).
 For Class 2 separations, one or more components appear in only one of the products.
 If neither distillate nor bottoms contain all feed components, two pinch points occur away
from the feed stage (figure d). Stages between the feed stage and the rectifying section
pinch point remove heavy components that do not appear in the distillate. Light
components that do not appear in the bottom product are removed by the stages between
feed stage and the stripping-section pinch point.
 If all the feed components appear in the bottoms, the stripping-section pinch point moves
to the feed stage (figure e). Alternatively if all feed components appear in the distillate, the
rectifying-section pinch point moves to the feed stage. 27
Minimum Reflux
 However, in multi-component system, there may be either one pinch point or
two pinch points. The multi-component systems are defined as of Class 1 and
Class 2 depending on pinch points.

 Only one pinch point can occur at feed point when all components are
distributed both in distillate as well as in bottom product. This is usually for a
mixture of components with narrow boiling point range.

 For a mixture with wide boiling point range there may occur two pinch points.

 Two pinch points occur away from feed points, when neither the distillate nor
the bottom product contain all the components.

 The pinch point in stripping section moves to feed plate if all components
appear in bottom products.

 The pinch point in rectifying section moves to feed plate if all components
appear in distillate.

28
Underwood Equation -
Minimum Reflux
 Binary systems
 pinch point usually occurs at the feed
plate

 Multicomponent systems
 separate pinch points in both
stripping and enriching sections if
there are nondistributing components

29
 Underwood Equation -
Minimum Reflux
 Underwood developed a procedure to find the
minimum reflux ratio for multi-component
systems

 Nondistributing light non-keys present

 a pinch-point will occur in the enriching section

 Nondistributing heavy non-keys present

 a pinch point will occur in the stripping section

30
Underwood Equation -
Minimum Reflux
 At minimum reflux Eq. 15.15:
c
 i Dx i,dist  i Bx i,bot 
Vmin - Vmin =  + 
i=1   i -   i - 

L min
 =
Vmin K HK
31
Underwood Equation -
Minimum Reflux
 To simplify:
Fzi = Dxi , dist + Bxi , bot

 i Fzi 
Vfeed = Vmin - Vmin =  
i  i - 

 1st Underwood equation

 used to calculate values of 
32
Underwood Equation -
Minimum Reflux

 Vfeed is the change in vapor flow

rate of the feed stage

 If q is known: Vfeed = F(1-q)

 If feed temp. is specified => use flash
calc. on feed to determine Vfeed

33
Underwood Equation -
Minimum Reflux

 i Dx i , disti 
Vmin =  
i   i - 

 2nd Underwood equation

 used to calculate Vmin

 Lmin = Vmin - D

34
 Underwood Equations – Cases
There are 3 cases that we need to consider when using the
Underwood equations:

Case 1: Non-keys do not distribute.

Case 2: Use the results from the Fenske equation and

assume that the distributions of non-keys
determined from Fenske equation are also valid at
minimum reflux. No intermediate key is present.

Case 3: All components distribute but no use of Fenske equation

is made. Exact solution without further assumptions.
35
 Underwood Equations –
Case 1 Approach
Case A: Assume all NK’s do not distribute – there are no intermediate NK’s.

1.) Specify the LK, HK, LNK’s and HNK’s.

2.) Determine q for the feed conditions.

3.) Determine all αi-ref‘s where the reference component is the HK.

4.) Determine ∆Vf from Eq.

ΔV f  F1  q 

5.) Solve Eq. (15.16) for a single root, φ, :

αi z i
ΔV f / F  1  q   Eq.(15.16)
i 1 α i  

Note that  LK-ref     HK-ref and choose the guess value between αLK and αHK.
If one chose the HK as the reference for the relative volatilities, αHK-ref = 1.
36
 Underwood Equations –
Case 1 Approach (continued)
6.) Determine the NK rates, Dx HNK i dist and Dx LNK i dist , from the assumption that
the NK’s
do not distribute:

Dx HNK i dist 0 Dx LNK i dist  Fz F, LNK i

7.) Determine the LK and HK rates from Eq. (15.5) :

Dx LK dist   f LK dist Fz F, LK

Dx HK dist  1   f HK bot FzF, HK Eq. (15-5)

8.) Determine Vmin from Eq. (15-13):

α i Dx i , D 
Vmin   Eq. (15-13)
i αi  

37
 Underwood Equations –
Case 1 Approach (continued)
9.) Determine D using Eq.:

D   Dx i, D 
i

10.) Determine Lmin and then Rmin from given Eq.:

L min  Vmin  D = (Rmin + 1) D

38
 Underwood Equations –
Case 2 Approach
Case B: Assume that the distribution of non-keys determined from the Fenske equation at total
reflux are also valid at minimum reflux.

1.) Specify the LK, HK, NK’s, LNK’s and HNK’s.

2.) Determine q for the feed conditions.

3.) Determine all αi-ref‘s where the reference component is the HK.

4.) Determine ∆Vf from Eq. :

ΔVf  F1  q 

5.) Solve Eq. (15.16) for one value of φ :

αi z i
1- q   Eq. (15.16)
i αi  

Note that  LK-ref     HK-ref and choose the guess value between αLK-ref and αHK-ref. If one chose
the HK as the reference for the relative volatilities, αHK-ref = 1.
39
Underwood Equations –
Case 2 Approach (continued)
6.) Determine the NK rates, Dx NK i dist , from the assumption that the distribution of NK’s non-
keys determined from the Fenske equation at total reflux are also valid at minimum reflux:

Dx NK i dist  Dx NK i dist, Nmin

7.) Determine the LK and HK rates from Eqs. (15.5):

Dx LK dist   f LK dist Fz F, LK

Dx HK dist  1   f HK bot Fz F, HK Eq. (15.5)

8.) Determine Vmin from Eq. (15.13):

α i Dx i , D 
Vmin   Eq. (15.13)
i αi  

40
 Underwood Equations –
Case B Approach (continued)
9.) Determine D using Eq.:

D   Dx i, D 
i

10.) Determine Lmin and then Rmin from given Eq.:

L min  Vmin  D = (Rmin + 1) D

41
 Underwood equations-
Case 3 Approach
 Assume all the components distribute but no use of Fenske
equation is made.
 Solve 1st Underwood equation for all values of lying 
between the relative volatilities of all components

 LNK , 1 > 1 > LNK , 2 >  2 >  LK >  3 >  HK > 4 >  HNK , 1

Give C-1 valid roots




 Write the Vmin eqn. for each value of

 Solve these eqns. for Vmin and Dxj,dist for all LNK and HNK
components by solving simultaneously (j represents non-key)

α i Dx i , D 
Vmin  
i αi  
42
Methods for using
Underwood equations -
Method 3

 Calculate average value of Vm

 Then find D and Lmin as usual

 Sandwich component problem must use

this approach
 more  's between LK and HK.
43
Using Underwood eqns.

 If only 1 root desired:

 LK >  >  HK

 Good 1st guess :

 = ( LK +  HK )/2
44
 Using Underwood
equations

 Results of Underwood equations will only be accurate

if the basic assumptions of constant relative volatility
and constant molar overflow are valid.

 Care must be taken to check if some of the

components do not distribute. Only those roots of eq.
15.16 must be considered to find Dxi,D whose values
lie between relative volatility of distributing
components. So, the use of case 3 may lead to
absurd results.
45
Theoretical stages for actual
reflux
 For binary mixtures for constant molar flow and
constant relative volatility N = f(ziF, xiD, xiB, q, R, )

For multi-component,
 Nmin = f(xiD, xiB, ) – Fenske’s equation

 Rmin = g(ziF, xiD, q, ) – Underwood equation

It was expected then that actual number of plates is a

function of
 N = N(Nmin, Rmin, R)

46
Gilliland Correlation

 Correlation for Number of Stages at

Finite Reflux Ratio

 Gilliland - emprically related N at finite L/D

to Nmin and (L/D)min

Eq. (15.18)
Where

47
Gilliland Correlation


Figure 15.4: Relation between reflux

ratio and number of theoretical stages
48
Gilliland Correlation

 Procedure to use the Gilliland

correlation:

1. Calculate Nmin from the Fenske

equation
2. Calculate Rmin or (L/D)min from Underwood
equation
3. Choose actual R or (L/D)
49
Procedure to use the Gilliland
correlation

4. Calculate abscissa
L
D -  
L
D min
=
 R - R min 
 
L
D + 1  R +1 

5. Determine ordinate value from

graph
 N - N min 
 
 N +1 
50
Procedure to use the Gilliland
correlation

6. Calculate actual number of

stages, N

Use correlation only for rough

estimates

51
 Estimate of optimum feed plate
location

1. Use Kirkbride equation to estimate

feed stage:

(Eq.15.20)

--Uppermost tray in the stripping

section will be the feed tray

52
Problem 1

53
Under wood Equation

54
Solution

55
Problem 2

56
Solution

57